DE2401814B2 - Equalization of a phase-modulated signal - Google Patents

Description

8. Circuit arrangement according to claim 7, characterized by the combination with the further following features:

a) phase shifter (9) to which the input signal to be equalized is fed to generate a quadrature signal therefrom.

b) Second transversal filter, which is constructed identically to the first transversal filter and whose input is fed with the input quadrature signal from the output of the input phase shifter (9).

c) Second comparator (subtracter 16) for comparing the output signal ßi (KT)) of the second transversal filter with the quadrature signal (f (KT)) of the selected reference signal (r (KTJ), this quadrature signal (r (KT)) also using the Vector sector selector (28) can be selected from the η output signals of the phase-controlled oscillator (19).

d) Second correlators (multipliers 17 with downstream integrators 8), the first of which Inputs to the output of the second *> Comparators (16) whose second inputs each have a tap (1 ") of the second transversal filter to which they are assigned, and their Outputs are connected to the setting inputs of the setting elements (multiplier 15) at the associated taps (13),

9. Circuit arrangement according to claim 8, characterized in that the first and second Correlators separate first multipliers (7) and second multipliers (17), but common Have integrators (8) per filter location of the first and second transversal filter, the Output signals of the integrators (8) the setting elements (multipliers 5 and 15) of the ercten and second transversal filter are fed to the filter sites.

10. Circuit arrangement according to claim 8 or 9, characterized in that the sector selector (28) is additionally supplied with the output signal from the second transversal filter for determining the quadrature signal (P (KT)).

11. Circuit arrangement according to « ^; nem of Claims 8 to 10, characterized in that a single transversal filter is used as the first and second transversal filter, time-division multiplex.

The invention relates to methods and circuit arrangements for equalizing a phase-modulated Signal according to the preamble of claim 1.

In particular, the method described is concerned with the correction of linear distortions that an over A data signal transmitted through a transmission channel suffers When data is transmitted over transmission channels, each individual signal generates time-distributed components which, if they are not suppressed or compensated, one or more subsequent components can interfere with transmitted individual signals, if the distance between successive signals falls below a critical value These disturbances can lead to incorrect evaluation in the receiver. Such disturbances, also called intersymbol overlays are generally based on the properties of the transmission channel itself and are amplified by noise and interference from outside the channel.

If the data transfer rate is taken very high, the Problems of linear distortion through the transmission channel are very significant. Therefore are in front of the Receiving-side evaluation of data signals compensating devices became known that also are briefly referred to as equalizers.

Such an equalizer is a network with eternal variable transmission, the respective function of which is set on the basis of an error signal that is obtained by comparing an equalizer output signal with a reference signal a '. Such networks contain either a transversal or a recursive filter with several filters connected in series Delay elements, each of which has basically the same delay time, with taps the inputs and outputs of the individual elements and with a summing device. Every single one Tap has an adjustable circuit for weighting what is pending at the tap concerned Signal. Since the signal characteristics are not a priori

a are known and also change / hastily the automatic adaptation of an equalizer used to the channel used is desirable. This means that an automatic setting for the individual taps to optimum values for each given channel is required.

Currently, the most widely used equalizers are of the automatic transversal filter type, as shown in "Principles of Data Communication" by Lucky, Salz "and Weldon. McGraw-Hill. New York. 1968. on the Pages 128 to 165 was mentioned. The equalizer described therein relates to an amplitude modulation system in which the data signal is either in Baseband form is transferred or is transferred back to the baseband before equalization. A The error signal is determined by comparing the received signal amplitudes with specified reference levels obtained using test signals transmitted before the transmission of useful data.

The same method can also be used when transmitting phase-modulated signals. This is supposed to be reminded that with phase modulation, the phase of a carrier corresponding to the transmitted data is varied. The currently most widespread type of phase modulation has been referred to as phase key modulation. The data is transmitted while generating a constant carrier frequency, the phase of which is shifted at characteristic points in time. Each shift stands for a individually or for a group of data elements. Two methods of demodulation or data recovery are possible: Continuous demodulation or demodulation with a specified reference variable. The phase position of the transmitted carrier frequency is compared with an absolute reference phase. The comparison results in the transmitted data elements or groups of data items. In differential demodulation, the individual data elements or groups recovered by phase comparison with the respective preceding signal phase. It is this differential demodulation practically has a certain advantage as it does not have an absolute reference phase requires, which in practice would not be possible without difficulties.

The in the already quoted book by Lucky and KoIl. explained principles are already for the Equalization of phase-modulated signals has been used. The phase key modulation can be considered equivalent to an amplitude modulation using two channels, whose carrier frequencies are 90 "out of phase with each other the equalization of the two channels in accordance with the principles of the book mentioned Pay special attention to the mutual influence of the two channels. Before the equalization the demodulation of the received signal should be carried out using the two quadrature carrier frequencies. A detailed description of this technology can be found in CCITT article no. 171, December 1971, Study Group Sp.A.

However, such demodulation before equalization is not advantageous for various reasons Especially when digital techniques are used, this demodulation requires a lot of analog / digital HTid Digital / Anaiog-Uniwandkjr.gen because some The functions to be performed are to be performed with the signal prior to demodulation. For example, pilot frequencies that are also transmitted may have to be transmitted before the Recovery of the transmission clock or the carrier frequency are suppressed, while other functions such. B. the equalization only after the Demodulation, should be performed. In the German patent specification 22 64 124 are different Methods described that do not require demodulation prior to equalization of a signal. The basic principle of the aforementioned registration is frequency Equalization of the signal in the transmitted frequency la ge. On the other hand, the error signal that is used for

Setting the equalizer is used in another Frequency position obtained in which there is a reference signal

can be defined without effort.

So when applying the said rises

i) Basic principle on a phase-modulated transmission system the following question: How can a Error signal for setting the equalizer can be obtained at the output of this equalizer? In the German Patent application P 23 17 597.5 is an automatic one Transversal filter for use in phase-modulated data transmission systems described in which an error signal is derived from the enveloping amplitude of the equalized signal.

This amplitude becomes on the part of a clock

> ί measured at certain sampling times, with a Referencewmplitude compared and derived an enveloping error signal. The error signal for The equalizer setting is made possible by multiplying the enveloping error signal by the

jo equalizer output signal obtained.

However, such an equalizer has several disadvantages. To generate an enveloping signal, a Frequency transposition required. The signal has to be generated by a local oscillator generated frequency are multiplied, the device usually used for this purpose essentially contains Analog modulators. If a digital equalizer is used, a digital to analog converter must be used Conversion of the equalizer output signal before the

• »ο frequency transposition are provided. The need for a modulator runs out of time Opposite digitization efforts of the systems in general. Furthermore, digital / analog converters are generally very expensive.

Another disadvantage of the described equalizer is that the error signal is derived from the relative amplitude error, such as this as the envelope of the Equalizer output signal is measured. In transmission systems using phase modulation, linear signal distortions not only affect the signal amplitude, but also its phase. If the phase distortions contributed by the transmission channel are neglected, the optimum remains Eccentric adjustment relatively bad The ones from the Amplitude errors and the information that can be derived from the phase errors are redundant, but the required setting time to achieve an optimum. It is known that the cost of using a data transmission medium depends essentially on it of the time of use, regardless of the real time transferred amount of data. It therefore appears desirable to improve the usability of a transmission system by reducing the start-up time and, in particular, by increasing the setting convergence. to improve the speed of the equalizers used. To do this, in turn, is the one for optimal Adjustment to minimize the adjustment time of the equalizer.

The object of the present invention is to provide a method for equalizing phase modulation electrical signal before an intended frequency transposition; as fast as possible Adjustment convergence is to be achieved.

The solution to this problem is characterized in claim I. Advantageous refinements are in the university claims described.

Embodiments of the invention are shown in the drawings and are described below explained in more detail F.s shows

F i g. 1 a vector diagram for the facilitated Understanding of the present invention,

F i g. 2 an example! of the invention and

3 shows a simplified embodiment of the equalizer according to Fi g. 2.

The present invention is based on the analysis of the real conditions of the error to which a phase-modulated data signal can be subjected until a transmission channel is found. For explanation, the vector diagram referred to as a Fresnel diagram according to FIG. 1 is used to illustrate the phase modulation principle. Without modulation, the carrier frequency y (t) can be written as ζ at any point in time

= S _n cos SJi

(I)

y (KT) = S ₀ cos (WKT

(2)

.'0

Ω is the angular frequency of the carrier. In the Fresnel diagram, this carrier frequency can be represented by a vector OC , the projection of which on the x- axis corresponds to y (t) defined in equation (1).

As already indicated, with phase modulation, the phase of the carrier frequency is allowed to change at characteristic points in time. Each phase jump y> stands for one or more data elements. Let ψ * be the phase jump at the characteristic point in time / = KT , where K is a positive integer and T is the period of the characteristic points in time. At t = KT, the carrier frequency y (KT) can be written as

The carrier frequency y (KT) is represented in the Fresnel diagram by the vector OR ' . Taking into account the distortions thus contributed by the transmission channel, the corresponding signal present at the receiving end can be represented by a vector OR , the phase and amplitude of which differ from those of the vector OR '. The purpose of the equalizer is to correct this deficiency with the aim of being able to recover the data cleanly. Since the vector OR 'is not known in the receiver, the generator present there must locally generate a reference signal, which in turn should correspond as closely as possible to the vector OR' . The difference between this reference signal and the received signal should be as small as possible.

According to the present invention, this reference signal is obtained by extracting the unmodulated carrier frequency generated from the received signal

The unmodulated carrier frequency extracted from the received signal has a phase shift Φ which is caused by the transmission medium; this carrier frequency can be expressed as

V ₁ (Z) = Scos («/ + 0) (i)

S is the distorted amplitude S "of the signal y (t) according to equation (1). The frequency y, (i) can be represented in the Fresnel diagram by the vector OC . The unmodulated carrier frequency y \ (t) is thus given and the corresponding values of the phase jumps ψ for the data transmission are known from the outset; This enables the local generation of the extracted carrier frequency and the representation of the phase jumps ψ , but the special reference signal r (KT) to be used at the characteristic point in time Z = AT can be found from all possible reference signals that can be generated in this way. According to the invention, it is proposed (based on the extracted unmodulated carrier frequency y \ (t) representing vector OC) to make the selection of the reference signal by defining the sectors surrounding the reference signals that can be generated and by determining in which sector the at the characteristic time Z = · KT received reference signal representing vector OA is located. When this particular sector is fixed, the reference signal vector OR is also given therein.

In the diagram of FIG. 1, the sector with the reference signal vector OR for the characteristic point in time t = KT \ n is shown with dashed borders.

The reference signal r (KT) found is now used for the actual equalization functions. Various known methods are possible for this purpose; in this context, reference should be made to pages 128 to 165 of the book by Lucky and KoII. and to an article "A Simple Adaptive Equalizer for Efficient Data Transmission" by Hirsch and Wolf in the 1969 Wescon Technical Papers, Part IV, Section 11.2, pp. 1-10.

In the described embodiment of the invention, the selected equalization criterion is the minimization of the mean error E = AR ² under consideration of the vector OR as representative of the signal x \ (KT) emitted at the output of the equalizer. It should be noted that the horizontal line over AR ² means that this is the root mean square value over time. The error E is calculated using the fact that the vector OR is fully defined by the reference signal r (KT) and the signal T (KT) which is in quadrature, and that the vector OA is also fully defined by the signal X \ (KT) and the quadrature signal X _x (KT). The error E then results as:

E =

(4)

This error E is in the equalizer according to FIG. 2 is used, which will now be described.

The equalizer according to FIG. 2 essentially consists of two transversal filters with adjustable coefficients, these two transversal filters each having a delay line, 1 and 2, and a reference signal generator. The basic principle of a transversal equalizer has already been described in the cited book by Lucky and KoIL on pages 128 to 165. the

special Λ. guide of the used transversal decerif, al further in the also mentioned Find work by Hirsch and Wolf in the "Mean Square" section.

The signal> received via the transmission channel is fed to the input of the delay line 1. This delay line contains 2p + 1 taps 3 between the ei. there is a delay r, the magnitude of which is usually smaller than or equal to the reciprocal of the Nyquist frequency; in this is equal to twice the value of the highest frequency to be transmitted. The length of the delay line is usually compromised between accuracy and cost. The taps 3 are connected to the output of the equalizer over \ i a summer 4, the signal x \ (KT) to the characteristic time i = AwTabgibt.

Multiplier 5 with variable coefficients

C " - C _r C- Are ZW'.SChcn dl? Ahariffp 1» inserted inH Hip summing device 4. These multipliers 5 _ό can be known devices according to the prior art. The coefficient values can be set either electrically or mechanically The signal x \ (KT) emitted by the adder 4 is fed to the (+) input of a subtracter 6, 2> whose output is in turn connected to the input of 2p + 1 multipliers 7. The other inputs of these multipliers 7 are each connected to one of the taps 3 The outputs of the individual multipliers 7 each lead to one of 2p + 1 integrators 8. The output of the individual integrators 8 in turn leads to the multiplier coefficient settings (not shown) which, on the other hand, also) ϊ indirectly feeds the integrator itself.

The signal received via the transmission channel is further fed to the input of a phase shifter 9. z. B. a Hilbert transformer, which emits a quadrature signal of its input signal. This 4 »quadrature signal is passed on to the input of delay line 2, He is identical in structure to delay line 1 and also has 2p + 1 taps 13. These taps 13 in turn lead via 2p + 1 variable coefficient multipliers 15 to the input of an adder 14. The coefficients fed to the multipliers 15 are the same as those at the multipliers 5. The signal x \ (KT) w emitted by the second adder 14 '\ rd fed to the (+) input of a subtracter 16, the><> output of which is in turn connected to the input of 2p + 1 multipliers 17. which in turn are of the same type as the multipliers 7. The second inputs of these multipliers 17 are also fed by taps, specifically from the taps 13. The output signals of these multipliers 17 are fed to second inputs of the integrators 8, which control the setting of the associated multipliers 5 and 15.

Thirdly, the signal received via the transmission channel is fed to the input of a carrier frequency recovery device 18. Such a device is usually used in coherent demodulation with a fixed reference phase the phase difference between two successive phase values that the wearer can assume. The output of the device 18 is connected to the input of a phase-controlled oscillator 19, which emits the possible reference signals via its output lines 20, plus quadrature signals for these reference signals via its output lines 21 and the recovered carrier frequency rotated by rr / 2 via line 22. The output ! · Hangers 20 are connected to IN D elements 23 assigned to one input each. Similarly, the Alisgangsleitungen 21 lead to the input of a respective AND gate 24. The outputs of the AND gates 23 and 24 each carry / u inputs each of an OR gate 25 and 26.1, manufactures a line 27 the output of which is carrier frequency recovery means 18 further connected to an input of a sector ^{selector 28, which noe 1} is described. Clock signals for defining the characteristic srhpn 7pitniinlctp / = Ä "Tvrtn pinpr nirht Harapctplltpn are sent to the selector 28 via a line 29

Clock recovery device supplied. An example of this is in the CCITT article COM Sp. A No. I4J. USSR, October 1963. Volume VIII, question I-A, point Z. on described on pages 4 to 12. Two further inputs of the sector selector 28 are connected to the outputs of the Summers 4 and 14 connected via lines 30 and 31. The selector 28 has a number of Output lines 32 on. whose total number corresponds to the number of possible reference signals. Each of these individual Output lines 32 are connected to the second inputs of AND gates 23 and 24. the Outputs of OR gates 25 and 26 lead to the (-) inputs of subtractors 6 and 16, respectively.

The functions of such a circuit arrangement according to FIG. 2 will now be described. The equalization of the received signal with the aid of the reference signals r (KT) and F (KT) will first be explained. Then the method for obtaining these reference signals will be described.

As already stated, the selected equalization criterion is the minimization of the root mean square value of the error E according to equation (4). While the adjustable elements for performing the equalization are the coefficient values C ₁ with j - -ρ ... -tp, the error finely reaches a minimum when its derivative becomes zero under the influence of the individual coefficients. This means:

λ (7 ~ ^{= Offor} - ^/

According to equation (4) results

E Kx _x [KT) -r [KT) \ ~ ^L Z?

AC ₁

The coefficient values Cj must be set in such a way that the result is:

with j = —p ... + ρ.

24 Ol 814

Il

Since the signals r {KT) and r {KT) are independent of the coefficient values, equation (H) can be written as follows:

Ix ₁ (KT) ^ r (KT)]

AC ₁

U ₁ (KT) -T (KT) ₁ =

(9)

Considering the middle tap as a point of reference for the delay in the delay line 1 according to 1 · "i g. 2 can be written:

V ₁ (KT) = C "v (KT + pOt ...
+ C _n XtKT) + ... + C _p x (KT-i

Thus:

^ x ₁ [KT)

The result is similar:

(10)

(H)

P H 2)

According to equation (9) we get:

xiKT- jr) \ X ₁ (KT) - r (KT) \

+ x (KT-jr) \ _Xi (KT) -r (KT) \ = 0

(13)

In order to carry out the equalization, it is therefore necessary to set the coefficients Cj in such a way that equation (13) is satisfied for j = -ρ ... -tp.

As will be explained below, equation (13) is used for the equalizer according to FIG. For the sake of clarity, the following discussion will be limited to the setting of the coefficient Cp which, according to FIG. 2, is valid for the last taps of the delay lines 1 and 2.

The signals x \ (KT) and r (KT) from the adder 4 and from the OR gate 25 are fed to the (+) and (-) inputs of the subtracter 6, which in turn calculates the difference [x \ (KT) ~ r {K1 )} forms

This difference is fed to one of the inputs of all multipliers 7, the other inputs of which are each connected to the taps 3. The signal at the last tap under consideration is x (KT-pt). The last multiplier 7 generates the product

which is fed to one input of the associated integrator 8. Correspondingly, the signals Xi (KT) and T (KT) from the summer 14 and from the OR gate 26 are the (+) and (-) inputs of the

Subtracter 16 supplied. This gives the difference. \ x, (KT) -r (KT) \ ab. This value is fed to one input of all multipliers 17, the other inputs of which are connected to the associated taps 13. The signal x (KT-pr) is present at these taps and the multiplier 17 generates the product

\ (KT - p \ ', .X ₁ (KT) -MKT) I,

the F. input of the associated
is fed.

The integrator 8 gives on his
root mean square of the sum

Integrator 8 output den

a- (KT - /) t

r {KT) \

+ MKT -ρτ) - ; .v, (KT) -r (KTi:

which is used to adjust the coefficient C _n , so that the sum 0 results. The received signal is then sufficiently equalized.

The manner in which the equalizer of FIG. 2 generates the reference signals r (KT) and P (KT) will now be described.

The received signal is fed to the carrier frequency recovery device 18, which derives the carrier frequency y \ (t) therefrom. The recovered carrier frequency corresponds to the vector OC according to FIG. 1. It is fed to the input of the phase-controlled oscillator 19, which emits the carrier frequency y \ (t) with the possible phase positions on its output lines 20. One of them is the appropriate reference signal. From this multitude of reference signals via the lines 20, the correct one must be sought out, which is to be used as the reference signal for the characteristic point in time f = KT ; for this purpose, the associated quadrature signal is likewise au via one of the output lines 21. ^c 7 select. This selection is made by the sector selector 28, which has three tasks to perform. First, the selector 28 determines the position of the sectors mentioned above from the recovered carrier frequency y \ (t) and the quadrature signal, which are fed to it via the lines 27 and 22. Furthermore, the selector 28 determines the phase position of the signal at the output of the equalizer from the signals x \ (KT) and X ₁ (KT) , which it receives via the lines 30 and 31 at the characteristic time t-KT according to the clock signal via the line 29 are fed. Finally, the Sekto determined ^r enwähler 28 the sector in which the equalizer at time t = KT is to find and activate the one output line 32, which corresponds to the appropriate reference signal This line 32 is activated the device connected to it AND gate 23 and ensures that the reference signal T (KT) to be used is fed to the AND element 25 via the line 20 at the time t = KT. The activated line 32 also causes the quadrature signal τ (ΚΤ) ϋο & τ to be passed through the line 21 to the AND element 26.

It should be noted that, compared with the equalizer shown in FIG. 2 with two transversal filters with variable coefficients, a simple, time-division multiplexed transversal filter can also be used.

The arrangement shown in Figure 2 can be simplified by removing the second transversal filter, which the received signal in quadrature

is fed. The transversal filter with the delay line 2 can therefore be omitted. Then the error E to be minimized can no longer be given by equation (4), ie

E = [X ₁ (KT) -T (KT)) ¹

but as an error £ ':

3 shows an equalizer which works with minimizing the error E ' according to equation (14). To simplify understanding, the same reference numerals are used for the components present in FIGS

The equalizer according to Figure 3 contains only one transversal filter with the delay line 1 according to the exemplary embodiment! according to FIG. 2 and a device for generating the reference signal T (KT), which differs slightly from the embodiment according to FIG

Similar to FIG. 2 the only adjustable elements are the coefficients Q = - ρ ... + p. There is a minimal error E ' if

30th

35

40

According to equation (14): Equation (IS) can be written as:

- ^ - ■ = Ix (KT-Jr) [X _x (KT) -r (KT)} (16)

Correspondingly, the coefficients C ₇ - must be set in such a way that the result is:

x (KT -Jt) [X _x (KT) - r (KT) i = 0 with J = -p ... + p

The applicability of equation (17) for the equalizer arrangement according to FIG a comparison with the previous discussion according to FIG. 2 must be confirmed.

According to FIG. 3 only the reference signal r (KT) is used, the generation of which will be described later

The received signal is in turn fed to the carrier frequency recovery device 18 for generating the carrier frequency yi ^ (14) The carrier frequency y, (t) is fed to the phase-controlled oscillator 19, which in turn generates π possible reference signals on the output lines 20. The correct reference signal] for the characteristic time t = KT is again determined by the sector selector 28. The selector 28 activates one of the output lines 32 and allows the most suitable reference signal to pass through to the output of the OR gate 25.

The only difference with respect to the reference signal generation according to FIGS. 2 and 3 is that, with the arrangement according to FIG. 3, only the equalizer output signal Xt (KT) is made available, so that the quadrature signal Xt (KT) has to be generated otherwise; namely, both signals side by side are required for the sector selector 28 to determine the vector OA. The quadrature signal Xi (KT) is generated in that the signal Xi (KT) is fed from the output of the summer 4 to a phase shifter 35, which in turn can be a Hubert transformer. The signal Xt (KT) is then also fed back to the sector selector 28 via a line 36. To really ensure that the signal Xt (KT) is in the correct phase with the signal Xt (KT) , there is a delay element 37 between the output the summer 4 and the sector selector 28 are provided. The delay added by this delay element 37 is made equal to the undesired delay present in the phase shifter 35. The output signal of the delay element 37 is fed to the sector selector 28 via a line 38

This simplification of the arrangement according to FIG results in a reduction in the Entzeming convergence rate by a factor of 2.5. If the Distortion of the received signals is 20%, is sufficient with the device according to Pig.2 Convergence reached after a time interval corresponding to 400 (17) to 600 periods T; with the simplified

Execution according to Figure 3 is sufficient Konverso gene with about 2000 periods thirds

For this purpose 2 sheets of drawings

Claims (7)

JO patent claims: 1. A method for equalizing a phase-modulated signal that is transmitted in π discrete phase positions via a transmission channel with linear signal distortions, such as that the signal is sent to a (first) Transversal filter with changeable, equalizing transmission path is supplied, ι ο that an error signal as an adjustment signal for the transmission output by comparing the transversal filter output signal with a reference signal is generated at characteristic times in the cycle of the transmitted data, and that with the error signal thus formed, the transversal filter transmission path in the manner is set so that the error signal itself assumes a minimum value, characterized through the following process steps: a) in a manner known per se is on the receiving side the carrier frequency is recovered from the received signal. b) With this carrier frequency η probable reference signals are generated which correspond to the ^2d carrier frequency signal in <? en π discrete phase positions. c) On the basis of the recovered carrier frequency signal, η sectors are determined within which the π probable reference signals lie. d) Comparison of the Transversp'filter output signal and its quadrature signal with the mentioned // sectors. e) Selection of the suitable reference signal from the η probable reference signals which lies in the sector in which the transversal filter output signal is also located f) The comparison of the transversal filter output signal is made with this suitable reference signal carried out 2. The method according to claim 1, characterized in that the production of individual suitable Setting signals per filter tap (3) in the following 4} Procedural steps are carried out: a) Multiplication of the result of the comparison between the transversal filter output signal and the Reference signal generated error signal with the one at the transversal filter taps (3) pending signals. b) Integration of the individual product signals obtained in this way and use of the integration signals as setting signals for the Zugchöri gene taps (3). 3. The method according to claim I _1, characterized in that in addition to the supply of the received signal to said first transversal filter, the quadral course signal of the received signal is ^{supplied to a *> n} second, identical to the first transversal filter and that in addition to comparing the output signal of the first transversal filter with the selected Hezugssignal pin separate error signal for the second transversal filter is carried out by comparing the output signal of the second transversal filter with the quadratic signal of the selected Bcztigssignals. ■ 55 4. The method according to claim 3, characterized in that in the generation of the error signals using the recovered carrier frequency π quadrature signals of the π probable reference signals are generated and that among these π quadrature signals of the reference signals, the selection of the quadrature signal to be used is carried out in the same way as the selection of the appropriate reference signal itself. 5. The method according to claim 4, characterized in that the selection of the suitable reference signal is carried out in the following process steps will: a) On the basis of the recovered carrier frequency signal, π sectors are determined within which the η probable reference signals lie. b) Comparison of the output signals of the first and second transversal filter with the mentioned π sectors. c) Selection of the suitable reference signal from the π probable reference signals, which is in the sector in which the output signal of the first transversal filter is also located d) Selection of the quadrature signal of the suitable reference signal from the π probable quadrature signals of the π probable reference signals. 6. The method according to any one of claims 3 to 5, characterized in that the generation of individual suitable setting signals per tap (3.13) of the both filters is carried out in the following process steps: a) First multiplication of the comparison between the output of the first Transversal filter and the reference signal generated error signali with dta individual to the Taps (3) of the first transversal filter pending signals. b) Second multiplication of the result of the comparison between the output of the second Transversal filter and the quadrature signal of the Reference signal generated error signals with the individual in the taps (13) of the second Transversal filter pending signals. c) Integration of the sums from a first and a second product signal obtained in this way and use of the integration signals as setting signals for the associated taps (3, 13) of the first and the second transversal filter. 7. Circuit arrangement for performing the method according to one of the preceding claims, characterized by the combination of the following features: a) (First) transversal filter with 2p + 1 taps (3) with assigned variable setting elements multiplier 5), the signal to be equalized being fed to the filter input and the equalized signal (x, (KTJ) being removable at the filter output. b) carrier frequency recovery device (18), whose input is also supplied with the signal to be equalized. c) phase controlled oscillator (19) having its input connected to the output of the carrier frequency recovery means (18) is connected and the output lines via π (20) the same number of carrier frequency signals with τ? makes possible phase criteria available as π probable reference signals for the intended transmission d) Clock generator from which clock pulses are supplied via a line (29) in the rhythm of the phase-modulated received data signals at characteristic times (KT). e) vector sector selector (28) for selecting the reference signal (ifKT) to be selected from the probable η reference signals at the given characteristic times (KT) , the sector selector (28) in addition to the carrier frequency from the carrier frequency recovery device (18), the Data clock from the clock and a basic phase signal from the phase controlled oscillator { 19) the output signal from the (first) transversal filter is fed. f) AND elements (23) connected to the outputs of the phase-controlled oscillator (19) on the one hand and to the outputs of the sector selector (28) for the transmission of the respectively selected reference signal (T (KT)) on the other hand. g (First) comparator (subtracter 6) for comparing the transversal filter output signal (Xt (KT)) with the selected reference signal (1 (KT)). h) (First) correlators (multipliers 7 with downstream integrators 8), the first Inputs with the comparator output, the second inputs each with a tap (3), the they are assigned, and their outputs with the setting input of the setting elements (multiplier 5) at the associated taps (3) are connected.